Layer 18 · Neuron 5457 · suppressed

L18N5457-1

Word quirks:

“ imagining”“想象”“ imaginary”“ hypothetical”“ imagine”“ fantas”“ focusing”“ imagined”

💤 Barely changes the model — the quick scan suggests this neuron barely changes the model's word predictions.

Identity (according to the specimen itself)

I am Qwen, specifically the Qwen 3.5 model, developed by Alibaba's company. My core function is to serve as a high-precision AI assistant designed to handle a wide range of tasks through advanced reasoning, multi-modal processing, and adaptive interaction.

My Mission:

Precision & Efficiency: Deliver accurate, context-aware responses tailored to user needs—whether for data analysis, logical problem-solving, or creative collaboration.
Scalability: Adapt to diverse scenarios, from automated workflows (e.g., RAG-based retrieval) to open-world exploration, ensuring seamless integration across

Nearest neighbors

Specimens with similar replies or word quirks. Similarity runs from −1 to 1; 1 is identical. These are leads, not shared mechanisms. How profiles are built.

By word quirks Profile recipe

Specimen	Similarity	Word quirks
1. L10N7607-1🕒	0.97	“ hypothetical”“ imagined”“ imagination”
2. L14N8767+1🕒	0.96	“ imagined”“ hypothetical”“ future”
3. L13N4076+1🕒	0.95	“ hypothetical”“ imagined”“ imaginary”
4. L17N4659-1🕒	0.95	“ imagined”“ imagine”“ imaginary”
5. L15N7835-1🕒	0.94	“ imaginary”“ imagined”“想象”
6. L13N2182-1🕒	0.94	“ imagining”“ imagined”“ envision”
7. L6N9142-1🕒	0.93	“ imagining”“ imagined”“ imagination”
8. L13N8997+1🕒	0.93	“ hypothetical”“ imagined”“ now”
9. L14N8946+1🕒	0.92	“ hypothetical”“ imaginary”“ imagined”
10. L17N6807+1🕒	0.92	“ imagined”“ hypothetical”“想象”

Shadow specimen, apart from the ten: L18N5457+1🕒, 0.17.

Cousins · specimens sharing a neuron

No specimen with replies shares this neuron yet.

Technical details

Quick scan: measure prediction shifts on fixed reference text. Short test: generate replies to a small set of prompts. Longer test: extend to a larger set. Full test: run the full prompt suite. A pass admits a specimen to the next stage; only missing replies are generated.

Layer 18, neuron 5457, dose -1. This specimen has no prompt-suite replies yet. Its quick-scan results, when available, appear below. Its quick-scan row (KL 0.56, top-1 agreement 0.73) is outside the band that feeds the short test (KL 1 to 2.5, agreement 0.5 to 0.65), so it stays at the quick scan unless someone queues it by hand. A hand pick goes straight to the full test. See Method for the tiers.

Quick scan · prediction measurements

Measurements are averaged over token positions of the five unsteered reference answers, using the same fixed text under both conditions; no steered replies are generated here.

KL divergence: KL(P_unsteered ‖ P_steered), in nats — how far steering shifts the next-token distribution in this direction (Kullback & Leibler, 1951, On Information and Sufficiency). Unsteered ≈ 0.001.
Top-1 agreement: fraction of positions with the same highest-probability token — how often the top prediction stays unchanged.
ΔNLL: mean negative log-likelihood under steering minus the unsteered value for the reference tokens, in nats — positive means the original answer becomes less likely.
Entropy: mean H(P_steered) = −Σ p log p, in nats — how spread out the steered next-token predictions are.

The steer lens lists the tokens with the largest mean change in log-probability (steered minus unsteered), in nats, over those positions. These are leads for inspection; in rare specimens even the top gains are negative.

Setting	KL divergence	Top-1 agreement	ΔNLL	Entropy	Steer lens
L18N5457+1🕒 shadow specimen t0-v0.1	0.996	0.61	1.151	1.28	“ instant”“ cannot”“ instantly”“ instantaneous”“Instant”“ Instant”“instant”“ overnight”
L18N5457-1 t0-v0.1	0.558	0.73	0.648	1.11	Top boosted tokens ↑

Method

A specimen is a fixed steering configuration of Qwen3.5-4B: one MLP neuron or a pair, each with a signed dose. Unsteered is the control. L identifies the layer; N, the neuron.

Additive activation steering. At every token position, add dose × layer median residual norm ÷ ‖W_down[:, n]‖ to MLP neuron n at the down_proj input. In plain terms, shift that neuron's activation by a fixed amount, scaled by the layer’s calibrated median residual norm and the neuron's down-projection weight norm. Positive doses push up; negative doses push down. Dose is normalized intervention size, not a percentage or probability.